IP layer mobility operation and beyond Keiichi Shima Internet - - PowerPoint PPT Presentation

▶
ip layer mobility operation and beyond
SMART_READER_LITE
LIVE PREVIEW

IP layer mobility operation and beyond Keiichi Shima Internet - - PowerPoint PPT Presentation

Sep 22, 2023 320 likes •865 views

IP layer mobility operation and beyond Keiichi Shima Internet Initiative Japan / NAIST The Second Asia Future Internet School PROJECT Topics global operation beyond the infrastructure-based mobility Mobile IPv6 experiment toward the

slide-1
SLIDE 1

IP layer mobility

  • peration and beyond

Keiichi Shima Internet Initiative Japan / NAIST The Second Asia Future Internet School

PROJECT
slide-2
SLIDE 2

Topics

  • Mobile IPv6 experiment toward the

global operation

  • Mobility technology application

beyond the infrastructure-based mobility

2

slide-3
SLIDE 3

Mobile IPv6

  • Network layer (Layer 3) mobility

protocol

  • On top of IPv6
  • Backward compatibility
  • Less impact to the existing infrastructure
  • Several extensions, e.g.
  • Network mobility
  • Dual-stack support

3

slide-4
SLIDE 4

Issues

  • Mobile IPv6 is a kind of a tunnel

based protocol

  • Single point of failure of the tunnel

server (home agent)

  • Redundant path (due to location of

mobile nodes and their home agents)

4

slide-5
SLIDE 5

Global operation

  • Single point of failure
  • Locate several home agents

around the world

  • Redundant path
  • Use nearest home agent

5

slide-6
SLIDE 6

Global HAHA Concept

Internet

  • The same route information is advertised to

the global Internet

  • Nearest agents will serve mobility requests

Home site 1 Home site 2 Home site 3 Mobile terminal

Home network route info Home network route info Home network route info

slide-7
SLIDE 7

How it works

  • Forward traffic goes to the nearest HA and forwarded to CN
  • Reverse traffic goes to the nearest HA (to CN) and tunneled

to the HA nearest to MN using the HAHA network

Home site 1 Home site 2 Home site 3 Mobile terminal

MN:HS1 MN:HS1 MN:HS1

7

slide-8
SLIDE 8

How it works

  • Home agent at home site 1 notices MN is now moved

to home site 3

  • Home agent sends a migration message to MN

Home site 1 Home site 2 Home site 3

MN:HS1 MN:HS1 MN:HS1 Change HS1 to HS3

slide-9
SLIDE 9

How it works

  • Forward traffic is now terminated by HS3
  • Reverse traffic is also forwarded by HS3

Home site 1 Home site 2 Home site 3

MN:HS3 MN:HS3 MN:HS3

9

slide-10
SLIDE 10

SHISA

  • IPv6 mobility development/research

infrastructure for BSD operating systems

  • Supported RFCs
  • Mobile IPv6 (RFC3775, 3776), NEMO BS

(RFC3946), Multiple CoA Registration, IPv4 traversal [experimental]

  • User space protocol signal processing
  • Easy to support new protocols
  • Kernel level packet forwarding
  • Keep forwarding performance

10

slide-11
SLIDE 11

Protocol Verification

  • We believe rough consensus and

working code, don’t we?

  • Verify the protocol by extending

SHISA framework

  • Using a real testbed

11

slide-12
SLIDE 12

Global Operation Implementation Design for SHISA

  • Only small modification is required
  • 1. HA to HA tunnels are established before operation

using the generic IP tunnel mechanism

  • 2. Binding cache information is copied using a newly

defined Mobility Header signal message

  • 3. HA switch message is sent using a newly defined

Mobility Header signal message

  • 4. Packet forwarding from HA to HA is implemented

using the standard host route mechanism

12

slide-13
SLIDE 13

Testbed

  • Interop Tokyo 2008
  • One of the biggest exhibition/conference for

network equipment/service vendors

slide-14
SLIDE 14

Topology

  • 2 home agents
  • 4 foreign networks
  • Home network route

information is advertised from two different locations by OSPFv3

ShowNet backbone Internet HA1 HA2 hall 4,5 hall 1,2,3 Server Segment Staff Segment Conference Segment Home network Home network CN MNb MNa

home network route info home network route info

14

slide-15
SLIDE 15

Verification Items

  • Check if a MN registers to the nearest

HA when booting

  • Check if a MN re-registers to a nearer

HA when it moves to the network close to the HA

  • Check if performance is improved by

changing HA

15

slide-16
SLIDE 16

MN6 MN5 MN4 MN3 MN2 MN1

11th June 2008 12:00 p.m. 12th June 2008 12:00 a.m. 12th June 2008 12:00 p.m. 13th June 2008 12:00 a.m. 13th June 2008 12:00 p.m.

On staff segment (2001:3e8:8::/64) On shownet (2001:3e8:b201:113::/64) On shownet123 (2001:3e8:b114::/64) Home agent switch message detected

Network near HA1 Network near HA2

16

slide-17
SLIDE 17

Results

  • MN could register to the nearest HA when

booting

  • MN could re-register to nearer HA if necessary

when it moves

  • We couldn’t verify performance enhancement
  • Because the network scale was too small to

check performance difference

  • We need larger scale testbed to confirm the

enhancement

17

slide-18
SLIDE 18

Other findings

  • Ping-pong registration

problem

  • We relied on the

underlying event network for packet routing

  • A load balancer

sometimes works as we do not expect

ShowNet backbone

HA1 HA2 Load Balancer

Registration Message Load balancer equally forwards the registration message to two possible routes

1.Mobile layer need to coordinate with underlaying routing layer 2.Load balancer must have knowledge of Mobility protocols

slide-19
SLIDE 19

Beyond the Infrastructure based mobility

slide-20
SLIDE 20

Background

  • Increasing threats of natural disasters

in urbanized cities

  • Increasing threats of artificial

disasters, like terrorism in crowded parts of a city

  • High risk to get into collapsed

structures

20

slide-21
SLIDE 21

Current Status

  • Remote rescue operation using

robots is intensively being researched

  • e.g. http://www.rescuesystem.org
  • Investigation of disaster areas using a

robot controlled by a human operator

21

slide-22
SLIDE 22
  • Ex. Crawler Robot
  • A robot with many

crawlers

  • Each crawler is

connected by a joint with high degree of freedom

  • Can get over
  • bstacles in

disaster areas

Photo by Matsuno Laboratory at the University of Electro-Communications, Japan

22

slide-23
SLIDE 23

Problems

  • Most of the robots are designed to be

controlled by a simple remote control method (e.g. with a wired remote)

  • The range that the robot can move around

is limited by the range of the remote

  • An operator must get into the disaster

area with the robot to control it, that may cause a secondary disaster

23

slide-24
SLIDE 24

Assumed Environment

  • Inside buildings (e.g. Subway

stations, underground malls)

  • Large searching area
  • No communication infrastructure
  • Unstable communication

environment

24

slide-25
SLIDE 25

New Network designed for Rescue Activity

  • Backbone is consists of multiple

wireless IP routers

  • Rescue robots will connect to the

nearest wireless IP router

  • A new wireless router is carried and

located to extend the network itself

25

slide-26
SLIDE 26

Operation Image

Point Zero Target

26

slide-27
SLIDE 27

Operation Image

Point Zero Target

26

slide-28
SLIDE 28

Operation Image

Point Zero Target

26

slide-29
SLIDE 29

Operation Image

Point Zero Target

26

slide-30
SLIDE 30

Operation Image

Point Zero Target

26

slide-31
SLIDE 31

Operation Image

Point Zero Target

26

slide-32
SLIDE 32

Operation Image

Point Zero Target

26

slide-33
SLIDE 33

Operation Image

Point Zero Target

26

slide-34
SLIDE 34

Requirements

  • Automatic network construction
  • Recovery from network failure
  • Data type based communication
  • Scalability

27

slide-35
SLIDE 35

Data Type Based Communication

  • Ad-hoc mesh network properties
  • Bandwidth changing time to time
  • Delay jitter
  • Unstable connectivity

28

slide-36
SLIDE 36

Data Types

  • Data transmitted over the network
  • Network management data
  • Robot remote control data
  • Sensor, image data

29

slide-37
SLIDE 37

Data Categories

  • Size
  • Acceptable delay time for each

type

  • Importance of data for each type

30

slide-38
SLIDE 38

Data Categories

Large data size Small data size Higher delay acceptable Lower delay required

Robot control Heartbeat Remote voice Remote camera Manipulation camera Topology Info Routing Info Sensor

slide-39
SLIDE 39

Selection of Data

  • Ideally, all communication should be
  • perable, however
  • Data selection is required based on

the environment

  • Sophisticated robot control
  • Network support

32

slide-40
SLIDE 40

For Practical Networks

  • Network technology doesn’t solve

all the problems

  • Coordination with robotics

technologies

  • Enhancement of UI technologies

33

slide-41
SLIDE 41

Networker’s Approach

  • As a part of the entire vision
  • Build APs with lower cost
  • Higher bandwidth
  • Manet based routing
  • Traffic control priority
  • Network as an application

34

slide-42
SLIDE 42

Multihop Wireless

  • Using cheap wireless technology

(IEEE802.11) and IP

AP2 AP3 AP4 AP5 AP6 AP7 AP1

Ch60 Ch64 Ch36 Ch40 Ch44 Ch48

35

slide-43
SLIDE 43

Multihop Wireless

  • Using cheap wireless technology

(IEEE802.11) and IP

AP2 AP3 AP4 AP5 AP6 AP7 AP1

Ch60 Ch64 Ch36 Ch40 Ch44 Ch48

35

slide-44
SLIDE 44

Terrible Results

  • UDP performance

measured by netperf

  • At 6 hops, only one-

fourth performance

  • f 1 hop case could

be achieved

0Mbps 5Mbps 10Mbps 15Mbps 20Mbps 1 2 3 4 5 6hops

36

slide-45
SLIDE 45

Interference

  • Wireless module

interfere each other even we use different channels

  • Direction antenna
  • Different bands

37

slide-46
SLIDE 46

Mixture of Bands

  • Direction antenna is

hard to operate

  • Using different bands
  • n each link as much

as possible

  • cf. “Routing in Multi-

Radio, Multi-Hop Wireless Mesh Networks” by MSR

0Mbps 7.5Mbps 15.0Mbps 22.5Mbps 30.0Mbps 1hop 2hops 3hops

38

slide-47
SLIDE 47

Overlay Manet

  • Manet protocols sometimes

depends the implementation design

  • f layer 2
  • Implement Manet protocol using
  • verlay L3 network

39

slide-48
SLIDE 48

Overlay Manet

  • Virtual Ethernet using

tun/tap interface

  • Manet protocols can

be implemented over tun/tap interface, as if they are operated

  • ver Ethernet
  • Detailed explanation

by Sho FUJITA, Tadashi YASUMOTO

Operating System (Linux, OSX, *BSD) TUN/TAP IFACE middleware Existing APPs DYMO SMF

  • 40
slide-49
SLIDE 49

Network as Application

  • Utilize the expanded

network as an information infrastructure

  • Each AP sends

snapshot image to robot operator

  • Each AP sends traffic

measurement data to monitor application

slide-50
SLIDE 50

Network is just one part

  • f the system

AP carrier Network status indicator

Robots Access Points User Interface

Environment camera Network monitor GIS Multiple robots

  • peration UI

Laser range finder Front camera Top and Rear camera

slide-51
SLIDE 51

Video

slide-52
SLIDE 52

Summary

  • Infrastructure based layer 3 mobility

technology completes its core parts

  • Next step is to establish the global

scale operation technology

  • We are proposing such a mechanism

and verifying it using our mobility infrastructure and with real neworks

44

slide-53
SLIDE 53

Summary

  • We have more frontier of mobility

research and development

  • Infrastructure-less environment (e.g.

disaster rescue)

  • Self-extensible network design and

implementation

  • Network as an information application
  • Integration with other core activities

45